Recent advances in molecular biology and mathematical modeling facilitate the genetic analysis of common human disorders which result from the interplay of multiple genetic and environmental factors. The necessity to correctly specify the genetic model in the usual maximum likelihood genetic linkage approach limits its application for the study of quantitative traits, and particularly in the study of multivariate phenotypes. This proposal builds upon previous methodological developments of less heavily model-dependent approaches for genetic analysis of complex disorders. We expand on our previous studies of the Haseman-Elston regression approach as well as a variance-components approach for genetic linkage. The proposed variance-components approach has some of the model-free properties of the Haseman-Elston approach, but more easily incorporates the family structure and permits the estimation of genetic parameters, as well as covariate and environmental effects. Thus, this variance component approach provides a semiparametric method for analysis of data from both population and family studies and provides a bridge for analysis of data collected according to usual molecular epidemiologic and genetic epidemiologic approaches. A multivariate extension of the Haseman-Elston approach has been developed but no simulation studies are available to provide guidelines for its use in small samples. The variance-components approach can also be extended to permit analysis of multivariate data, and simulation studies will be used to evaluate the efficiency of estimation under either approach. Simulation studies will also be used to evaluate the comparative power of standard maximum likelihood linkage approaches with Haseman-Elston and variance components procedures. The variance-components approaches and, in a more limited fashion, usual maximum likelihood methods will be applied to data that have already been collected from studies of cardiovascular and gallbladder disease risk factors. Programs for analysis of data will be distributed freely.
Showing the most recent 10 out of 24 publications
